On Sat, Apr 21, 2018 at 11:11:52AM +0200, Tobias Hansen wrote: > thanks for the bug report. We are aware that brial has a circular dependency > with sagemath. src:sagemath build-depends on src:brial and python-brial > should depend on sagemath. We didn't add this dependency in order not to > manifest the circular dependency in the package relationships. We brought > this up with upstream once (can't find it now) but they didn't care. For that > reason I don't know if there is anything to be done at the moment.
Avoiding the circular dependency is a good thing. Let me suggest some ways to improve the situation anyway: * python-brial could at least Recommends: sagemath. Doing so wouldn't manifest in Build-Depend and thus does not constitute a cycle. * If sagemath uses python-brial for testing only (I cannot tell), then you can annotate the build dependency with <!nocheck> introducing a better way to break the cycle. If that works, I think having the cycle in the normal case is a better trade-off. * Sometimes it is feasible to move build dependencies to Build-Depends-Indep. That can also be used as a cycle breaker. * Since brial does not Build-Depends: sagemath, is there actually a cycle? Or can you simply go and add the dependency. Helmut
